Spectromorphological analysis of sound

objects: an adaptation of Pierre Schaeffers

typomorphology
LASSE THORESEN wi th the assi stance of ANDREAS HEDMAN
The Norwegian Academy of Music, P.O. Box 5190, Majorstua, 0302 Oslo, Norway
E-mail: lasse.thoresen@nmh.no
The last fifty years has witnessed an enormous development
with regard to sound production, and has opened a new world
of novel aural experiences. In order to be able to articulate
and discuss these experiences there is a need for a
corresponding novel set of terms and concepts. Such a
terminology would also be relevant for analytical and
interpretive approaches to electroacoustic music, avant-garde
Western music, and ethno music. Pierre Schaeffers
typomorphology, developed in the 1960s, proposed a variety
of novel terms, but they have not been of widespread use,
since they unfortunately did not lend themselves very well for
practical analysis. The present paper intends to develop
Schaeffers approach in the direction of a practical tool for
conceptualising and notating sound quality. While carefully
reducing the sometimes-bewildering number of terms found in
Schaeffers work, it introduces a set of graphic symbols apt
for transcribing electroacoustic music in a concise score. The
analysis of sound objects calls for a specific listeners
intention, called reductive listening.
1. INTRODUCTION
In so far as analysis is concerned, Western musicology
has focused on pitch structures (harmony, modality,
etc.), construction of musical forms (themes, motives,
etc.), and rhythm (metre). Timbre was traditionally
regarded simply as a matter of colourisation of musical
structure, and was treated in terms of orchestration.
The electroacoustic music, whose first breakthrough
was the musique concre`te of Pierre Schaeffer and Pierre
Henry during the late 1940s, profoundly challenged the
prevailing understanding of timbre. The developments
in technology and physics that took place during
subsequent years enabled humans to mould timbre
dimensions and thus to include these as an integral part
of a compositional design. The numerous and novel
aural experiences of sound made by composers and
technicians working in the studio called for documenta-
tion and theoretical as well as philosophical reflection,
so as to counterbalance the proliferating number of
engineering concepts that seemed to monopolise the
discourse of electroacoustic music. Undoubtedly, the
most substantial response to this challenge has been the
work on spectromorphology
1
by Pierre Schaeffer whose
Traite des objets musicaux (hereafter referred to as
TOM) appeared in 1966. It addressed the question of
correlation between the world of acoustics and engi-
neering with that of the listener. While the technology he
used for his experiments is long since outdated, the
overall perspective of the book is still valid. In his
approach to listening he provides a number of new
categories and concepts that are eminently suited to
discuss timbre and sound quality. In consideration of
the novelty of its subject matter, the originality of its
approach, and its depth of philosophical reflection, it
certainly deserves to be regarded as one of the most
important theoretical works of twentieth-century musi-
cal thought.
Schaeffers approach to the world of sound is
characterised by a phenomenological attitude: it seeks
to describe and reflect upon experience, rather than
explain; it posits the actual life world experience of
sound as its primary object of research (la primaute de
loreille); it clarifies a number of different listening
intentions by which the same physical object may be
constituted as various objects in the listeners mind. The
capacity to shift between different listening intentions
becomes a true sign of the virtuoso listener, and
Schaeffer insists that the listener should train his
listening even as a musician would train his instrument!
Unfortunately, and for a number of reasons, one of
the major achievements of Schaeffers work, his
codification of all sound categories into a grand, unified
diagram, remained without much practical conse-
quence. Through my teaching of Schaeffers typomor-
phology at the Norwegian Academy of Music since the
late 1970s, I have compiled a number of ideas about how
Schaeffers typomorphology could be made into a better
1
The term spectromorphology is certainly the most adequate
English term to designate the vast field of research opened by
Pierre Schaeffer and INA/GRM. It was coined by Denis Smalley.
He explains the thoughts that went into the creation of the term
thus: Lorsque jai commence (en 1981) a` elaborer un cadre (un
syste`me) qui me permette detudier le contenu de la musique
acousmatique sur la base des idees de Traite de Schaeffer, jai forge
le terme de spectromorphologie pour representer lidee des
composantes du spectre sonore la matie`re sonore et le domaine
des hauteurs et celle de leur evolution dans le temps leur
morphologie collective. Le terme combine donc les notions
schaefferiennes de matie`re et de forme. [] Je ne voulais pas non
plus utiliser le terme schaefferien de typo-morphologie, car il
nest pas toujours approprie de se referer a` un type (Smalley
1999: 183).
Organised Sound 12(2): 129141 2007 Cambridge University Press. Printed in the United Kingdom. doi: 10.1017/S1355771807001793
tool for practical analysis. The present paper is a
condensation of these experiences; it will present a set of
conceptual and graphic tools for the aural analysis of
music with an enriched sonic morphology. I have
profited greatly from the study of Michel Chions
indispensable work Guide des objets sonores. For the
revisions of Schaeffers ideas as well as the digitalisation
of the graphic signs, I am grateful for the assistance of
Andreas Hedman.
2
The terminology and analytical
notation proposed here is solely intended as one out of
many ingredients in what would eventually be a fully
developed aural analysis of music in general, and
electroacoustic music in particular. Music can, in
principle, be analysed on three levels: that of sound
objects, that of elementary patterns (i.e. organisations,
structures) combining sound objects, and that of
patterns of patterns (that last one being an analysis of
musical form). Music theory has generally concentrated
on the second level, taking the first one as granted. We
have, through the years in a project entitled Aural
Sonology, developed a series of approaches to the aural
analysis of level three in particular, by developing means
of analysing the music by subdividing it into units
(successive segments or simultaneous layers) as well as
the characterisation of different functional relationships
between these (Thoresen 1985a, b, 1987, 1996). Later a
ground-breaking academic work has been done by
developing a methodical approach to the analysis of
electroacoustic music by Stephane Roy (Roy 2003). His
approach is in many ways parallel to ours. He too finds
Schaefferian terms useful in order to characterise basic
units(hepreferstocall theelementsonlevel oneunits, not
sound objects or musical objects). However, for a more
detailed analysis of this aspect of music (level one), our
approach offers a greater number of terms and signs.
3
2. THE ADAPTATION OF SCHAEFFERS
TYPOMORPHOLOGY TO PRACTICAL
ANALYSIS
Schaeffers ideas on the categorisation of sound objects
were summarised in TOM in the diagram called
TARSOM (Tableau recapitulatif du solfe`ge des objets
musicaux). This diagram sums up a number of other
diagrammatic representations, of which TARTYP
(Tableau recapitulatif de la typologie) is the most
important, as it is intended to present a presumably
all-encompassing typology of sound objects. Schaeffer
and Reibel later illustrated a number of these categories
with sound examples in Solfe`ge des objets sonores
(Schaeffer and Reibel 1966b). Here letters are used to
designate the different analytical categories of
TARTYP (see Figure 1). Incidentally, Schaeffer makes
a few suggestions about how these letters can be
combined into small chains in order to describe more
complex sound objects in an analytical context.
4
The
TARTYP imposes a normative view on the sound
objects. Consistent with Schaeffers intention to identify
musical objects fit to serve as elements of structure in a
new, utopian music, he labels some objects as suitable
(convenable), others as too unpredictable (trop origi-
naux), or too redundant (trop redondantes). However,
although one entire book within TOM is devoted to
the relationship between object and structure, the
question of structure, or more precisely, what the nature
of musical structures is, is left open. He does, however,
mention the need to develop a general musicology,
focused on structure (Schaeffer 1966a: 3501).
Schaeffers approach to schematisation is original in
its emphasis on relativity: it does not insist that every
sound object ought to have a corresponding unique
location in the diagram. For instance, one sound may be
characterised as a pitched object (son tonique), but since
it is a glissando it will be a variable, pitched object (son
variable), and since it is an element of an accumulation, it
could also fit into that category. This principle may be
unusual in a scientific context, but is all the more
reasonable from an artistic point of view. The relativity
is made acceptable largely because of the strength of the
two main axes that organise TARTYP: one is related to
the energy articulation (entretien) of the sound object;
the other to the character of sound spectrum or sonic
substance (masse). A third dimension, that of duration,
enters the diagram as well, as there are three categories
of durations that serve to discriminate categories: The
short impulse (micro-objets, dure reduite), the medium
range duration (dure mesure), and the excessively long
durations (macro-objets, dure demesure).
Beyond TARTYP, there are a number of distinctions
made and summarised in TARSOM, categorised under
the columns classes, genres and espe`ces. These are
multiplied through the seven criteria of musical percep-
tion: sound spectrum, dynamics, harmonic timbre,
melodic profile, profile of the sound spectrum, grain,
and gait. In most cases, Schaeffer does not indicate any
notation, such as e.g. a letter code that could facilitate
the use of his categories in practical analysis.
Without going into a detailed discussion of some of
the problems encountered whentrying to use Schaeffers
immense, and partly speculative system in practical
analysis of electroacoustic music (or of bird song, or of
ethnic music), let us briefly summarise the reasoning
behind the proposed revision of Schaeffers original
design.
2
A specially designed font, called Sonova, has been used to
produce the graphic signs used in the diagrams.
3
Roy goes ontodiscuss criteriafor dividingapiece intounits, applying
explicit criteria defined by principles taken from Gestalt Theory, in
combination with the paradigmatic method developed by N. Ruwet.
Inthis respect his approachis exemplaryinits explicitness. Heends up
with a little more than forty functions designed to characterise the
relationships between the sounds. However, since this article only
deals with the primary level of sound objects, it would not be relevant
to carry on with a detailed comparison of the two methods.
4
Formules Typologiques (Schaeffer 1966: 4667).
130 Lasse Thoresen
The most important feature of this revision is the
introduction of graphic symbols as opposed to letters or
verbal designations to represent the analysis. This opens
a number of new possibilities. Graphic symbols make a
multidimensional representation possible. Consider-
ations of typology and morphology may enter into an
integrated, compact representation of the sound object.
The disadvantage of using letters, as Schaeffer did, is
that they only lend themselves to the formation of
strings of arbitrary symbols, whereas graphic symbols
can be combined and also used to give iconic
representations of the sonic dimensions such as pitch/
register, durations, and the superposition of simulta-
neous elements. Graphic symbols can be drawn in
detailed ways that eliminate the need for a number of
diagrammatic categories (e.g. all categories dealing with
melodic profiles can be simply drawn, as can many
aspects of duration, pitch and register).
The introduction of a graphic notation calls for a re-
examination of the need to maintain all twenty-eight
categories of TARTYP; the present approach starts by
reducing it to its cardinal points. Moreover, the
normative dimension of TARTYP, i.e. the distinction
between suitable objects (objets convenables) and unsui-
table ones (objets trop originaux, objets trop redondantes)
is removed. Accordingly, the distinction between
facture and entretien, which also implied a qualitative
evaluation of sounds, is removed as well.
5
Consider-
ations of duration were removedfromthe diagram, with
the exception of the point zero of the energy axis, which
had to be maintained. Categories of duration were then
reintroduced by a subsidiary graphic notation that can
be added to the core symbols whenever desirable.
Accordingly, the redundant objects were removed
from the diagram altogether.
What now remains on the horizontal axis, that of the
energy articulation, is the logical line of moving from a
short impulse, via mediumdurations with simple objects
towards increasingly complex, ultimately unpredictable
objects. In keeping with Schaeffers original design, this
happens symmetrically from the middle.
On the vertical axis, the same basic distinctions are
preserved: pitched, complex (i.e. unpitched) and vari-
able sound spectra. We now have a minimal representa-
tion of TARTYP, consisting of the nine central
categories, as well as the two times three categories on
the extreme left and right, i.e. altogether fifteen cardinal
cases. What needs to be done next is to fill the voids
between the inner nine and outer six categories with
transitional objects. Once the diagram is reconstructed,
we will also see how the Schaefferian objects that were
removed can be regained, through the use of the graphic
tools, and listed as special cases.
The graphic symbolism developed is very well suited
for a digital representation. Andreas Hedman, a
Swedish composer associated with the EMS in
Stockholm, has produced a font that contains all the
Figure 1. TARTYP.
5
Facture is not aneutral termlike entretien. It is qualitative andimplies
that certainsoundtypes are more likely tohave musical potential than
others due totheir pitch content and dynamic shapes. Thus all sounds
have entretien, only some have facture (Dack 1998: 88).
Spectromorphological analysis of sound objects 131
elements needed to draw the standard symbols in a
simple text file.
2.1. Listening intentions and questions of pertinence
The Schaefferian project has made it eminently clear
that a persons aural perception of any object is founded
in one among many possible listening intentions. Thus,
there is an adequate listening intention corresponding to
each pertinent feature of the musical discourse. Any
verbal exchange about music as heard will tend to
become incongruent, even meaningless, unless there is
consensus regarding listening intentions and their
correlation to a stratum of pertinent sonic information.
The question of pertinence is, of course, delicate and
somewhat controversial when it comes to the aesthetic
appreciation of electroacoustic music (Delalande 1998).
On the other hand, it can be said that the lack of any
conventions with regard to discussing and analysing
electroacoustic music is detrimental both to pedagogy,
humanistic research, and possibly also to public
comprehension and acceptance of this novel art form.
The present article presents a set of analytical
conventions that presuppose the practice of a reductive
listening. This intention is characterised by the intention
to hear the sound simply as a sound, mentally bracketing
its indexical associations (ideas about the sound source),
as well as its traditional position in pre-existing musical
languages, and refraining from any other interpretation
of symbolic or semantic nature. Michel Chion sum-
marises the Schaefferian thoughts on the subject
succinctly:
The reductive listening is a listening attitude that consists in
listening to the sound itself, as a sound object, while
abstractingit fromits real orsupposedcause, as well as from
the meaning it might convey. More precisely, it consists in
turning this double curiosity for causes andmeanings (both
of which treat the sound as an intermediary between other
objects towards which it directs the attention) towards the
sound object itself (Chion 1983: 33)
This is a listening intention that easily can be taught and
shared, and is thus fit for providing one intersubjective
basis of observation. Complementary intentions may
and should of course be developed; I myself have for
instance been carrying out research on the aural
constitution of structure, in a post-Schaefferian spirit
for many years (see References). The method of analysis
presented here is designed in such a way that it can be
integrated into the still broader context of graphical,
analytical tools for describing aural thought.
Mastering the intentionality of open reductive listen-
ing is a first step that leads to careful observation of the
different attributes of the sound as such. It soon
becomes evident that the traditional terminology of
Western music theory does not contain the vocabulary
to discuss the emergent qualities of the sound objects.
Schaeffers typomorphology comes as a result of an
activity of predication: names have been assigned to a
number of different sound qualities, and a number of
different criteria of listener dimensions in single sounds
have been systematised. When sounds are listened to
with the purpose of placing them into a pre-existing
category, the openness of the reductive listening is easily
lost in favour of a more selective attention. Very quickly
we may impose conceptual prejudices on perceptual
givens. This may be an inevitable disadvantage of any
attempt to codify aural phenomena. However, once one
is made conscious of it, one can choose whether to
practice the open reductive listening or the categorising
reductive listening. Each of these attitudes are valuable
in their own right: the open reductive listening tending
towards the Husserlian epoche can often lead to the
discovery of newaspects of an object and leadfurther on
towards the creation of new musical ideas. However,
many observations in this realm of thought will be
impossible to communicate through words, and this is
where some shared, conceptual conventions will be of
use.
Reductive listening goes well with the taxonomical
approachtolistening, i.e. listeningtothe waythe music is
ordered (what are the units the piece falls into? What are
their relationships? etc.) Neither of these are sponta-
neous,layman waysoflistening.However, toamusician
and a composer, it is a necessary and required profes-
sional capability to master these listening intentions.
6
We shall now discuss in more detail the reorganisa-
tion of Schaeffers typological and morphologic con-
cepts, not so much to question their validity, but rather
with a view to their feasibility as practical tools of one
particular brand of aural analysis, termed spectro-
morphological analysis.
7
6
In his above-mentioned article on music analysis and reception
behaviours, Francois Delalande outlines a few listening behaviours.
Taxonomic Listening, Empathetic Listening, Figurativisationfeature
prominently in his article; inadditionhe proposes Searchof a Lawof
Organization, Immersed Listening, and Non-listening. Denis
Smalley points to the possible dangers of too much emphasis on
reductive listening. It is as dangerous as it is useful for tworeasons.
Firstly, once one has discovered an aural interest in the more detailed
spectromorphological features, it becomes very difficult torestore the
extrinsic threads to their rightful place. Secondly, microscopic
perceptual scanning tends to highlight less pertinent, low-level,
intrinsic detail such that the composer-listener can easily focus too
much on background at the expense of foreground. Therefore, while
the focal changes permitted by repetition have the advantage of
encouraging deeper exploration, they also cause perceptual distor-
tions. My experience of teaching composers has often revealed to me
that such distortions are frequent (Smalley 1997: 11).
7
The analysis of sound based on reductive listening is that aspect of
musical analysis that best would render itself for an automatic
analysis: a computer might analyse the physical aspect of sound and
link the result of the analysis up against an interpretation into
spectromorphological categories. Although this could be useful, e.g.
in case one needs a rough score, the overall objective of the process
of analysis, in our context, is the training of the aural consciousness
itself. The repeated listening to the sound and the effort to
determine its characteristics bring about a clearer aural awareness
of the anatomy of different sounds. The resulting interiorisation of
sonic qualities and their orientation in an overall conceptual
structure is a prerequisite for an intuitive, creative mental process.
132 Lasse Thoresen
3. DETAILED PRESENTATION OF THE
REVISED TYPOLOGY
The minimal representation of the typology (Figure 2)
shows only cardinal cases the extremities of the
organising axes that later on will serve to orient the
expandedversion of the scheme. The vertical axis sets up
three criteria of the sound spectrum (left hand side), the
horizontal axis deals with that of energy articulation.
The criterion sound spectrum is a definition of that
aspect of the sound in which the perception of pitch and
pitch content is founded. The sounds that have a clearly
perceivable pitch or fundamental will be termed pitched
sound objects (sons toniques). The ones with no
perceivable fundamental (drum sounds, tam-tam
sounds, wind, consonants, etc.) will be termed complex
or unpitched sound objects (sons complexes). Sound
objects with a gradual internal development in its sound
spectrum (glissandi or sounds with gliding formants)
will be termed variable sound objects (sons varies). These
may be either pitched or unpitched.
Beginning with the impulse (short thrust of energy)
and moving to the left, the impulse is prolonged and
comes to form sustained objects. Towards the right, the
object is prolonged by means of iteration, i.e. quick
repetitions as in a tremolando. On the extreme left of the
diagram we find sound objects that, although basically
sustained or continuous in energy, have an unpredic-
tably diversified energy articulation, and could accord-
ingly be termed vacillating sound objects (the English
term is not a translation of the French echantillon [lit.
Sample]). The creaking of a door, the cracking of the
tone produced by a badly handled bow on a string
instrument are examples of vacillating sounds.
On the extreme right we find sound objects called
accumulations. They are thought of as being over-
articulated iterations; i.e. iterations in which the
iteration pulse as well as the sound spectrum of the
single occurrences are unpredictable in detail. Examples
of these objects would be the sound of raindrops on a tin
roof, the sound of a flock of sparrows, or of peas
running out of a bag and hitting a table.
Interestingly, the extremes of the diagram meet:
vacillating sound objects and accumulations can be very
similar. Both of the types of sound objects can combine
pitched and complex sonic elements. One could speak of
homogenous accumulations (using only one type of
sound) as opposed to heterogeneous accumulations
(mixing different types of sound, particularly those with
different characteristics of sound spectrum). A similar
distinction can be made with regard to vacillating sound
objects.
3.1. The expanded typological diagram
In the simplified diagram two blank columns were left
open for transitional categories after we removed the
two categories of macro objects. In the expanded
diagram some additional categories are introduced
(Figure 3). Between pitched and complex sounds a
category termed dystonic sound objects (corresponding
to Pierre Schaeffers sons cannele) is inserted (Chion
1983: 1468). These are ambiguous sounds whose sound
spectrumis formedby a mixture of pitched elements and
clusters. Instruments like gongs, triangles, and bells fall
into this category.
All objects listed in the central three columns have
been doubled: there are both filled and empty note
heads.
N
The empty, round note heads will be used for
sinusoidal sound objects (sons purs).
Figure 2. Typology minimal representation.
Spectromorphological analysis of sound objects 133
N
The open diamonds will be used for whisper-like,
quasi-pitched sounds.
N
Empty square note heads will be used for
unvoiced complex sound objects (thus hhh-
sounds, broad bandwidths of white noise, etc., as
opposed to a filled square for drum sounds, clicks
etc.).
The intermediary category placed in the column on the
right diagrammatic axis will be termed composite sound
objects (these would correspond to a Schaefferian
category in his external morphology, called compose,
i.e. distinct and successive elements; Schaeffer 1966a:
4646; Chion 1983: 140). This category contains several
subcategories, moving from relatively simple cases
towards increasingly complicated objects. The simpler
cases would, for example, consist of a pair of sounds,
such as a grace note added to (or inserted into) a main
note, or a trill or mordent. Moving towards greater
unpredictability, we can add more notes, let them be
more diversified in sound spectrum, and/or let the
iteration pulse become increasingly irregular. I see no
reason to create a detailed conceptual structure contain-
ing various subcategories of composite objects, as the
character of the composite object in question will be
adequately rendered by an appropriate combination of
graphic symbols. However, the need arises for a
notational tool to describe degrees of irregularity, and
such a tool will be presented shortly.
The open void on the left side of the diagram is filled
with sounds called stratified objects (these would
correspond to a Schaefferian category in his external
morphology called composite, i.e. objects with distinct
and simultaneous elements; Schaeffer 1966a: 4646;
Chion 1983: 140). These are sustained sounds that show
different degrees of internal differentiation. The simplest
case is a sustained sound with a prominent harmonic.
From there on we proceed to increased spectral
differentiation, inclusion of elements of contrasting
sound spectrum, eventually spectral fluctuations with
increasingly irregular pulse, until we arrive at the
vacillating object. As was the case with the other
intermediary category, the exact nature of the object
may be shown through the graphic notation.
4. TYPOLOGIES OF DURATION AND OF
REGULARITY
As the consideration of duration was removed from
Schaeffers original design, and the concept of a gradual
transition between different degrees of regularity was
introduced into the intermediary categories, some
additional signs and definitions will be needed. I have
earlier worked out an analysis of types of velocities,
features of whichwill nowbe selectedfor inclusion inthe
spectromorphological analysis.
4.1. Types of velocity and duration
N
Gesture time. The central category of velocity/
duration is equivalent to the durations character-
istic of Schaeffers tenues formes. These are sounds
long enough to have an onset phase, a sustained
part and an ending, while short enough to be
integrated aurally into strings of sounds that can
be appreciated as a whole.
Figure 3. Typology expanded diagram.
134 Lasse Thoresen
N
Ambient time. To the category of very long/slow
durations belong those sounds whose duration is
so long that their sustained part dominates the
opening and ending phases disproportionately.
The durations of this category correspond to the
duration of Schaeffers homoge`nes. These are
sounds that would make the aural comprehension
of the relationship between more than two
adjoining objects difficult as they cross beyond
the retention/protention limits of immediate per-
ception.
N
Flutter time. When elements in a string of events
run so fast that they tend to integrate or become
blurred, we have reached the opposite end of the
scale of velocities and durations. To the extent
elements integrate, they become iterated objects,
which is a case already covered in the typology
diagram.
N
Ripple time. Between flutter time and gesture time
there is a transitional region, consisting of medium
fast strings of sound objects discernible one by one,
but tending to merge easily. In relation to a basic
pulse in gesture time, ripple time forms the
upbeats.
4.2. Pulse categories
Under pulse categories we will deal with phenomena
such as regularity/irregularity, tendential changes of
speed, etc. For the purpose of brevity, we shall confine
this discussion to the main cases, all of which are first of
all applicable to pulses of gesture time and ripple time.
We discern three degrees of periodicity:
N
Regular pulse divides time into equally long
segments, or in equal number multiples of
segments.
N
Irregular pulse divides time into unpredictable
durations.
N
Oblique pulse forms an intermediary category
between the former two. It may be based on
duration ratios like 3:2, 5:3, or come about
through the superposition of regular pulses.
Moreover, pulses may also change tendentially, i.e.
through accelerandi and ritardandi. The tendential
changes of time have been combined with indications
of the three degrees of periodicity into tremolando-like
graphic signs.
5. SPECIAL CASES
A number of cases that Schaeffer originally included in
his TARTYP have temporarily been excluded from the
central diagrams. Now, however, they can easily be
regained by combining already defined graphic sym-
bols. These are special case objects that can be named
and listed, but have not been assigned single, specific
graphic signs:
N
Sound web (trame): an object in ambient time with
constantly changing spectrum. A special case of
stratified sounds.
N
Large note (grosse note): an object in slow gesture
time with a slowly and predictably evolving
spectrum. A special case of stratified sounds with
variable sound spectrum.
N
Ostinato (pedal): a repeated sequence of sonic
objects, ostinato-like. A special case of composite
sound objects.
N
Cell (cellule): an accumulation with a total
duration in the range of gesture time.
N
Incidents: a special case of composite objects.
N
Accidents: a special case of stratified objects.
N
Homogenous sound (homoge`ne): a sound object
with stable sound spectrum, without evolution in
energy articulation, and of ambient time duration.
Figure 4. Types of velocity and duration. Figure 5. Pulse categories.
Spectromorphological analysis of sound objects 135
N
Fragment: an extremely short sound. Can be
notated by using the sign for the impulse half size.
N
Chord: a superposition of several sound objects of
similar kind.
N
Additionally, the vacillating sound objects in
gesture time, a case not incorporated in
TARTYP, can easily be represented by the graphic
tools.
6. MORPHOLOGY
Whereas the typology tried to sort out basic types of
sound and separate them one from the other, we will
now present more detailed criteria for how the anatomy
of a single sound can be described. In doing so, we have
taken a selection from Schaeffers prolific TARSOM in
order to make a workable tool for practical analysis.
6.1. Criterion: sound spectrum
N
Spectral width (classes de texture de masse) (Chion
1983: 146). The width of the spectrum is defined in
relationship to the extremities of sinusoidal sounds
and white noise (Figure 7). The diagram begins
with the sinusoidal shape, and then goes to the
pitched sound with an overtone spectrum, espe-
cially a harmonic one. The diagram then bifur-
cates, suggesting that the intermediary stages to
obtain a dystonic sound could pass through either
a chord of pitched sounds (with a suitable interval
structure that the notation does not specify) or
through a further saturation of the spectrum. This
logic is now followed further, to suggest the
transition between dystonic and complex sound.
N
Spectral brightness. An analysis of this dimension
is actually not included in Schaeffers typomor-
phology. The phenomenon is, however, well
known from linguistics: the vowel sound [i] is
considered brighter than [u] and intermediary
cases can easily be conceived. A similar scale can
be made for the complex impulse sounds of
language: [t] is, for example, brighter than [d] or
[g]. The characterisation of spectral brightness
may, for instance, be important for discerning the
colour difference between different instruments
that all produce pitched sound objects. The
graphic notation will vary according to the
stability or variability of the sound in question:
for stable brightness one sign will be used; for
continuously varying brightness, a sign character-
istic of stratified, variable sounds will be used.
Figure 6. Schaefferian special cases. Figure 7. Spectral width.
136 Lasse Thoresen
The vertical line put onthe prolongationline indicates
the entire spectrum from high to low. The small vertical
line on the left indicates the brightness of the spectrum,
fromhigh to low. The vertical line is ideally conceived of
as absolute, so that, for example, its lower region is
always the bass region, irrespective of the register of the
pitch to which this particular colour is predicated.
For sounds with variable brightness, the notation of
stratified sounds will generally be preferred. When the
open note head is put in parenthesis, it indicates that the
corresponding partial is not perceived as a separate
pitch or entity, but is an integral property of the sound
spectrum (its formant in acoustical terms). Spectral
profile will be the term used for the trajectory of internal
variations in the width of the sound spectrum of the
sound, e.g. the gradual transformation of a pitched
sound toa complex sound. Internal changes in the width
of the spectrum could be described as expanding,
convex, concave, or receding.
6.2. Criterion: dynamic profile
The dynamic profile of a sound object is intimately
connected to its energy articulation. Some of these
profiles are already implied in the typology diagram,
others are found in the list of special cases presented
above. Schaeffer differentiates the following dynamic
profiles:
8
N
No dynamic profile (dynamique nulle; no variation).
Typical object: the homogenous sound the static
sound object in ambient time.
N
Weak dynamic profile (dynamique faible; probably
an undulation in slow gesture time with oblique,
non-repetitive pulse). Typical object: sound web.
N
Formed dynamic profile (dynamique formee; a
profile in gesture time suggesting a beginning,
middle and end). Typical object: the balanced
objects in the two columns on each side of the
middle one of Figure 1.
N
Impulse-like dynamic profile (dynamique-impulsion;
characterised by a sudden thrust and decline of
energy, with no sustained phase). Typical object:
impulse.
N
Cyclic dynamic profile (cyclique; repetitive
dynamics). Typical object: ostinato.
N
Vacillating dynamic profile (reitere; irregular pulse
in a continuous energy flow). Typical object:
vacillating sound objects.
N
Accumulation-like dynamic profile (accumule; irre-
gular pulse in a discontinuous energy flow).
Typical object: accumulation.
There are dynamic profiles that are characteristic of
certain timbres, namely sounds of the type attack-
resonance. The attack or onset phase determines in fact
the character and duration of the resonant phase of the
sound. The ear is therefore particularly sensitive to the
timbre information given in the onset phase, and uses
this informationto identify its timbre andits source. The
information in the onset phase can be conceived of as a
bundle containing a certain characteristic duration,
8
See TARSOM, rubric 21; Chion (1983: 155).
Figure 8. Spectral brightness.
Figure 9. Spectral profile.
Spectromorphological analysis of sound objects 137
articulation and sound spectrum. Figure 10 shows the
proposal for onset phase genres; they differ in minor
respects from those of Schaeffer (Chion 1983: 158):
N
Brusque onset: e.g. a sound caused by a metal
hammer stroke against metal; the attack transient
is clearly and separately perceived.
N
Sharp onset: e.g. a sound caused by a mallet or
plectrum; the onset sound is closer to the character
of the main body of the sound as opposed to the
former case.
N
Marked onset: normal (non legato) onset of a wind
or string instrument.
N
Flat onset: no particular marking of the onset as,
for example, in legato playing.
N
Swelled onset: a short crescendo/decrescendo.
N
Gradual onset: the sound begins with a swell or a
crescendo.
N
Without onset: the onset cannot be heard dal
niente.
If desired, the onset can be further characterised by
adding an indication of the spectral brightness of the
opening transient (Figure 11).
A typology of onset phases ought to be complemen-
ted by a corresponding set of typical ways of ending a
sound. Such a typology does not exist in Schaeffers
TOM, or in his Solfe`ge. The reason may be that the
beginning of a sound naturally attracts more attention
than its ending; the opening phase of the sound often
contains its structurally pertinent features. There are,
however, cases where the way a sound ends can have
musical pertinence (e.g. when resonating sounds are
contrasted with sounds whose resonance is suddenly
interrupted).
Figure 12a shows the proposal for a typology of
ending genres. Since the ending of a sound generally
may be left unspecified, the majority of these categories
will concern sounds with a clearly audible ending
transient. It should also be pointed out that in the case
of impulses, there will be no need to describe the ending
phase at all. Thus the genres below will only deal with
prolonged sounds, either sustained or iterated.
Moreover, we have left out endings that are voluntarily
imposed on the sound, as they can conveniently be
designated through conventional dynamic signs.
N
Abrupt ending: sounds with an accentuated ending
phase, containing elements alien to the main body
of the sound; e.g. vibrating string dampened with
metal rod.
N
Sharp ending: sounds with an ending phase
containing elements intrinsic to the main body of
sound itself; e.g. harpsichord tone.
N
Marked ending: sounds stopped suddenly, rather
than rounded off.
N
Flat ending: unmarked.
N
Soft ending: the ending of the sustained sound is
rounded off with a diminuendo.
N
Resonating ending: a resonating sound is let free to
resonate until it dies out; laissez vibrer.
The sign for resonance will be a slur. The slur can be
combined with a prolongation line; the latter will then
indicate the length of the sound (e.g. bell sound that
vibrates after the attack). When the slur is added after
the prolongation line, it will either mean a laissez vibrer,
or simply suggest that the sound has a reverberation,
and that the duration of the sound is not indicated
precisely by the prolongation line. The case of marked
damping of a resonating body, the interrupted reso-
nance, is an important special case.
Abrupt, sharp or marked endings will occasionally be
reinforced by a crescendo, giving amongst others the
case of the reversed sound (e.g. a vibraphone sound
replayed backwards) (Figure 12b).
6.3. Criterion: gait
The term gait is an attempted translation of the French
word allure, meaning a way to walk. The English term
allure is unsuitable as a translation of the French word,
as it means to entice by charm or attraction (Infopedia
Dictionary). The word gait seems to render the French
Figure 10. Dynamic profile.
Figure 11. Spectral brightness of the opening transient.
138 Lasse Thoresen
better, as it means: a manner of walking or moving on
foot; a sequence of foot movements (as a walk, trot,
pace, or canter) by which a horse or a dog moves
forward (Infopedia Dictionary)
The gait of a sound object is the undulating move-
ment or characteristic fluctuation that often can be
found in the sustained part of sound objects. The gait of
a sound could be defined as being the idea of a vibrato
generalised (Chion 1983: 158). These undulating move-
ments can be traced in the pitch dimension, in the
dynamic dimension or in the spectrum of the sound
object. Thus we will distinguish between pitch gait,
dynamic gait, and spectral gait.
The gait will be further analysed with regard to its
degree of deviation (i.e. the degree of its departure from
an average value) and its pulse velocity (which will
generally range fromgesture time to ripple time; in order
to specify the latter we will add the tremolando-like
signs shown in Figure 5 over the prolongation line of the
gait sign). The design of the diagram is meant to
emphasise the five cardinal cases (extremes and middle
position) these would often be sufficient for practical
purposes.
Sometimes, the analyst will be facedwith the choice of
whether to represent what he or she hears as an
undulating glissando (using the main prolongation line
of the sound) or a slow and wide pitch gait. Generally,
gait is perceived to be more ornamental than a
glissando; a secondary, perhaps expressive musical
element, rather than a line with its intrinsic importance
for the musical discourse. The gait of a sound can be
seen as the signature of its source. The person behind
the voice, the individual player behind the violin tone
can be identified by the characteristic way in which the
sound is vibrating. If one generalises all soundsources to
three overall categories, namely living ones (pre-
eminently human), natural ones (i.e. phenomena of
nature), and mechanical ones (i.e. produced by
machines), it would appear that living sources have a
tendency to vibrate in a slightly oblique pulse, the
natural ones in an irregular pulse, and the mechanical
ones in a regular pulse (Chion 1983: 159).
Figure 12b. Ending genres.
Figure 13a. Pitch gait.
Figure 13b. Dynamic gait.
Figure 13c. Spectral gait.
Figure 12a. Ending genres.
Spectromorphological analysis of sound objects 139
6.4. Criterion: granularity
The microstructure of a sound object can be more or less
coarse or slick. The consideration of this aspect of the
sound will fall under the category of granularity. It is a
dimension of sound comparable to the abrasiveness one
can feel when touching a piece of cloth or a mineral, or
the granular quality one can discern in a photograph.
Granularity, then, can be perceived through three sense
modalities: sight, touch and hearing. For all of these
fields it can be described in the same way, i.e. as the
overall perception of irregularities of detail (grains)
that affect the surface of the object (Chion 1983: 152).
However, sometimes the distinction between granu-
larity and iteration can be hard to differentiate when
conducting an analysis. Generally, grains are a micro
feature of the object in question, whereas iterations are
of a coarser kind; thus grains would tend to be smaller,
quicker, and be inseparable from the main body of the
sound. A subsidiary consideration would be to regard
the idea of the cause of the grains; this is a slight
aberration with regards to the reductive listening, but
nevertheless one which Pierre Schaeffer himself pro-
posed when suggesting a typology of grains that
distinguished between grains resulting from beating
(such as the grains of the deep notes of the double
bassoon), rubbing (as when the violinist increases the
pressure of his bow), and resonance (such as the myriads
of small particles that can be imagined to exist in the
sound of a cymbal) (Schaeffer 1966a: 1524; Chion
1983: 5515).
A simplified representation of Schaeffers many
distinctions of granularity is shown in Figure 14a. The
coordinates of the diagram are Coarseness of the grains,
and its velocity, which in all cases would be in the region
of quick flutter time. A further analysis of the
phenomenon of granularity would look into:
N
the sound spectrum of the grain (to the extent it
differs from the sound spectrum of the carrier
sound),
N
the weight of the grain (how prominent the grain is
in relation to the carrier sound), and
N
the placement of the grain (i.e. in which register the
grain can be found).
Figure 14b shows the graphic notation of the latter
categories.
Figure 14a. Granularity.
Figure 14b. Granularity: spectrum, weight, placement. Figure 15. General additional conventions of notation.
140 Lasse Thoresen
REFERENCES
Chion, M. 1983. Guide des objets sonores. INA, Editions
Buchet/Chastel.
Dack, J. 1998. Strategies in the analysis of Karheinz
Stockhausens Kontakte fu r elektronische Kla nge,
Klavier und Schlagzeug. Journal of New Music Research
27(12): 84119.
Delalande, F. 1998. Music analysis and reception behaviours:
Sommeil by Pierre Henry. Journal of New Music Research
27(12): 1366.
Roy, S. 2003. Lanalyse des musiques electroacoustiques:
Mode`les et propositions, Preface de Jean-Jacques Nattiez.
LHarmattan.
Schaeffer, P. 1966a. Traite des objets musicaux. Editions du
Seuil.
Schaeffer, P., and Reibel, G. 1966b. Solfe`ge de lobjet sonore.
New editions: 1988, 2005.
Smalley, D. 1997. Spectromorphology: explaining sound-
shapes. Organised Sound 2.
Smalley, D. 1999. E
`
tablissement de cadres relationnels pour
lanalyse de la musique postschaefferienne. Ou r, entendre,
ecouter, comprendre apre`s Schaeffer. INA, Editions
Buchet/Chastel.
Thoresen, L. 1979. Hastighetstyper (Nor.) (Types of Velocity).
Unpublished Library of the Norwegian Academy of
Music.
Thoresen, L. 1985a. Un model danalyse auditive. Analyse
Musicale 1 (French translation of article published in
Semiotica).
Thoresen, L. 1985b. Auditive analysis of musical structures. a
summary of analytical terms, graphical signs and defini-
tions. InPro-c. fromICEMConf. on Electroacoustic Music.
Stockholm, Sweden, 2527 September.
Thoresen, L. 1987. An auditive analysis of Schuberts Piano
Sonata Op. 42. Semiotica 66(1/3).
Thoresen, L. 1996. Flerskiktighet i Olav Anton Thommessens
orkestermusikk. In Olav Anton Thommessen, Inspirator,
Tradisjonsbrer, Modernist. Norsk Musikkforlag.
Spectromorphological analysis of sound objects 141